CN107562758A - Information pushing method and device and electronic equipment - Google Patents

Information pushing method and device and electronic equipment Download PDF

Info

Publication number
CN107562758A
CN107562758A CN201610513018.0A CN201610513018A CN107562758A CN 107562758 A CN107562758 A CN 107562758A CN 201610513018 A CN201610513018 A CN 201610513018A CN 107562758 A CN107562758 A CN 107562758A
Authority
CN
China
Prior art keywords
characteristic
information
vector
pushed
msub
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201610513018.0A
Other languages
Chinese (zh)
Other versions
CN107562758B (en
Inventor
邓少伟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Kingsoft Internet Security Software Co Ltd
Original Assignee
Beijing Kingsoft Internet Security Software Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Kingsoft Internet Security Software Co Ltd filed Critical Beijing Kingsoft Internet Security Software Co Ltd
Priority to CN201610513018.0A priority Critical patent/CN107562758B/en
Publication of CN107562758A publication Critical patent/CN107562758A/en
Application granted granted Critical
Publication of CN107562758B publication Critical patent/CN107562758B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Abstract

The embodiment of the invention discloses a method and a device for pushing information and electronic equipment, relates to the internet information pushing technology, and can improve the information pushing quality. The information pushing method comprises the following steps: learning by using a decomposition machine to obtain learning characteristic vectors in a plurality of preset time periods, wherein each learning characteristic vector is a characteristic of information to be pushed; acquiring browsing statistical vectors of the information to be pushed in the multiple time periods; calculating a feature statistical vector of each dimension of features contained in the information to be pushed according to the learning feature vector and the browsing statistical vector; calculating the preference value of the target to the information to be pushed according to the feature statistical vector and the learning feature vector of the target; and if the calculated preference value exceeds a preset preference threshold value, pushing the information to be pushed to the target. The invention is suitable for article pushing.

Description

The method, apparatus and electronic equipment of a kind of pushed information
Technical field
The present invention relates to internet information push technology, more particularly to a kind of method, apparatus of pushed information and electronics to set It is standby.
Background technology
With electronic equipment, for example, intelligent mobile phone, personal digital assistant, palm PC, notebook computer etc. are applied It is more and more extensive, in order to lift the application experience for the user for using electronic equipment, by collecting user and associated user to going through The navigation patterns of history information predict the hobby of the user, and the information of user preferences is met to user's push, not only can be matched User's request, the valued added applications business of related operators can also be extended, realizes the win-win of user and operator.
At present, the information to be pushed content that focuses on of information push matches with user preferences, general using based on letter Collaborative filtering recommending (Collaborative Filtering recommendation) method of content is ceased, i.e., user is carried out User characteristics is classified, for example, classified according to age of user, sex etc., and the pushed information browsed to sorted users is carried out Keyword extraction simultaneously counts, and obtains navigation patterns of such user to each keyword, so as to build user characteristics and the information content Mapping relations；Then, according to user's classification belonging to user to be pushed, the keyword included in information to be pushed is extracted, is obtained Historical viewings behavior of the family to the keyword of extraction is taken, is learnt with reference to the mapping relations of structure, so as to judge the user To the hobby of the information to be pushed.Wherein, disassembler (FM, Factorization Machines) is the one of collaborative filtering Kind, by the way that information to be pushed to be decomposed into the information vector for including the implicit vector of multidimensional, targeted customer is decomposed into and information Dimension identical user vector, within the period pre-set, for example, one day, using study method to information vector with And user vector carries out assignment, then, dot-product operation is carried out to the user vector and information vector of assignment, user can be obtained and existed To the hobby of the information in the period.
A kind of but information recommendation method, because Factorization Machines are implicitly learned methods, after study Obtained each dimensional feature vector all without definite implication, thus, the data twice of different time sections input are learnt Afterwards, the characteristic vector non-equivalence obtained, cause the user that different time sections obtain larger to the hobby deviation of information, for example, according to According to the user for being learnt yesterday to obtain to the hobby of information and according to the user for being learnt today to obtain to information It is larger to like difference so that the information of the push hobby matching degree actual with user is relatively low, reduces information push quality.Example Such as, the information of push is not browsed or directly deleted by user by user.
The content of the invention
In view of this, the embodiment of the present invention provides a kind of method, apparatus and electronic equipment of pushed information, can lift letter Breath push quality, to solve in the method for existing pushed information, foundation disassembler is learnt the obtained different time cycle Hobby difference it is larger, information push it is of low quality the problem of.
In a first aspect, the embodiment of the present invention provides a kind of method of pushed information, including：
Learnt using disassembler, the learning characteristic vector in the multiple time cycles pre-set, each study is special Sign vector answers the feature in information to be pushed；
Obtain the information to be pushed in the multiple time cycle browses statistical vector；
According to the learning characteristic it is vectorial and it is described browse statistical vector, calculate included in the information to be pushed it is every The characteristic statisticses vector of one-dimensional characteristic；
According to characteristic statisticses vector and the learning characteristic vector of target, calculate the target and treat push letter to described The hobby value of breath；
If the hobby value calculated exceedes the hobby threshold value pre-set, treat that push is believed to described in target push Breath.
With reference in a first aspect, in the first embodiment of first aspect, using following formula calculate the characteristic statisticses to Amount：
In formula,
StatiFor the characteristic statisticses vector of the i-th dimension feature included in the information to be pushed；
frejWithin the multiple time cycle user of the information to be pushed is browsed for jth position user statistics to Amount；
featurejiFor jth position user within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed to Amount；
N is the number of users of statistics.
With reference in a first aspect, in second of embodiment of first aspect, using following formula calculate the characteristic statisticses to Amount：
In formula,
ζjiFor i-th dimension characteristic vector coefficient corresponding to the user of jth position；
ζjFor the characteristic statisticses vector coefficient of jth position user.
With reference to the first embodiment or second of embodiment of first aspect, in the third embodiment party of first aspect In formula, hobby value of the target to the information to be pushed is calculated using following formula：
In formula,
ψ is the hobby value of target information to be pushed；
featureuiFor target within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed vector；
K is the intrinsic dimensionality included in information to be pushed.
With reference in a first aspect, in the 4th kind of embodiment of first aspect, described calculate is wrapped in the information to be pushed The characteristic statisticses vector of the every one-dimensional characteristic contained includes：
The characteristic statisticses vector and characteristic information of information to be pushed in the acquisition cycle very first time are simultaneously stored with first Area stores；
Within the second time cycle, the characteristic statisticses vector and characteristic information obtained in the cycle very first time is carried out more Newly and with the second memory block store, and, the characteristic statisticses vector of the information to be pushed in the second time cycle of acquisition and spy Reference ceases and uses the first memory block to store；
Within the 3rd time cycle, the characteristic statisticses vector and characteristic information obtained in the cycle very first time is carried out more Newly and with the 3rd memory block store, the characteristic statisticses vector and characteristic information obtained in the second time cycle is updated simultaneously Stored with the second memory block, and, the characteristic statisticses vector and feature for obtaining the information to be pushed in the 3rd time cycle are believed Cease and use the first memory block to store；
Using the characteristic statisticses obtained in the cycle very first time of renewal vector as the information to be pushed being calculated In include every one-dimensional characteristic characteristic statisticses vector；
Within the 4th time cycle, the 3rd memory block is emptied, the feature to being obtained in the cycle very first time is performed and unites The step of meter vector and characteristic information are updated and use the 3rd memory block to store.
Second aspect, the embodiment of the present invention provide a kind of device of pushed information, including：Learning characteristic vector obtains mould Block, statistical vector acquisition module, characteristic statisticses vector calculation module, hobby value computing module and pushing module are browsed, wherein,
Learning characteristic vector acquisition module, for being learnt using disassembler, in the multiple time cycles pre-set Learning characteristic vector, each learning characteristic vector answers the feature in information to be pushed；
Statistical vector acquisition module is browsed, for obtaining browsing for the information to be pushed in the multiple time cycle Statistical vector；
Characteristic statisticses vector calculation module, for according to the learning characteristic it is vectorial and it is described browse statistical vector, meter Calculate the characteristic statisticses vector of the every one-dimensional characteristic included in the information to be pushed；
Hobby value computing module, for the learning characteristic vector according to characteristic statisticses vector and target, calculate institute State hobby value of the target to the information to be pushed；
Pushing module, if the hobby value calculated exceedes the hobby threshold value pre-set, institute is pushed to the target State information to be pushed.
With reference to second aspect, in the first embodiment of second aspect, using following formula calculate the characteristic statisticses to Amount：
In formula,
StatiFor the characteristic statisticses vector of the i-th dimension feature included in the information to be pushed；
frejWithin the multiple time cycle user of the information to be pushed is browsed for jth position user statistics to Amount；
featurejiFor jth position user within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed to Amount；
N is the number of users of statistics.
With reference to second aspect, in second of embodiment of second aspect, using following formula calculate the characteristic statisticses to Amount：
In formula,
ζjiFor i-th dimension characteristic vector coefficient corresponding to the user of jth position；
ζjFor the characteristic statisticses vector coefficient of jth position user.
With reference to the first embodiment or second of embodiment of second aspect, in the third embodiment party of second aspect In formula, hobby value of the target to the information to be pushed is calculated using following formula：
In formula,
ψ is hobby value of the target to information to be pushed；
featureuiFor target within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed vector；
K is the intrinsic dimensionality included in information to be pushed.
With reference to second aspect, in the 4th kind of embodiment of second aspect, the characteristic statisticses vector calculation module bag Include：Very first time period treatment unit, the second time cycle processing unit, the 3rd time cycle processing unit, characteristic statisticses to Acquiring unit and rotation unit are measured, wherein,
Very first time period treatment unit, for obtaining the characteristic statisticses vector of the information to be pushed in the cycle very first time And characteristic information and with the first memory block store；
Second time cycle processing unit, within the second time cycle, to the feature obtained in the cycle very first time Statistical vector and characteristic information are updated and use the second memory block to store, and, obtain and wait to push away in the second time cycle Deliver letters breath characteristic statisticses vector and characteristic information and with the first memory block storage；
3rd time cycle processing unit, within the 3rd time cycle, to the feature obtained in the cycle very first time Statistical vector and characteristic information are updated and use the 3rd memory block to store, to the characteristic statisticses obtained in the second time cycle Vector and characteristic information are updated and use the second memory block to store, and, obtain in the 3rd time cycle and treat push letter The characteristic statisticses vector and characteristic information of breath are simultaneously stored with the first memory block；
Characteristic statisticses vector acquiring unit, for the week very first time that will be updated in the 3rd time cycle processing unit The characteristic statisticses vector obtained in phase is united as the feature of the every one-dimensional characteristic included in the information to be pushed being calculated Meter vector；
Unit is rotated, within the 4th time cycle, emptying the 3rd memory block, notifies the 3rd time cycle to handle Unit performs the characteristic statisticses vector and characteristic information to being obtained in the cycle very first time and is updated and uses the 3rd to deposit The step of storage area stores.
The third aspect, the embodiment of the present invention provide a kind of electronic equipment, and the electronic equipment includes：Housing, processor, deposit Reservoir, circuit board and power circuit, wherein, circuit board is placed in the interior volume that housing surrounds, and processor and memory are set On circuit boards；Power circuit, for each circuit or the device power supply for above-mentioned electronic equipment；Memory is used to store and can hold Line program code；The executable program code that processor is stored by reading in memory is run and executable program code pair The program answered, the method for performing foregoing any described pushed information.
The method, apparatus and electronic equipment of a kind of pushed information provided in an embodiment of the present invention, by using disassembler Practise, the learning characteristic vector in the multiple time cycles pre-set, each learning characteristic vector is answered in information to be pushed A feature；Obtain the information to be pushed in the multiple time cycle browses statistical vector；It is special according to the study Levy it is vectorial and it is described browse statistical vector, calculate the characteristic statisticses of the every one-dimensional characteristic included in the information to be pushed to Amount；According to characteristic statisticses vector and the learning characteristic vector of target, the target is calculated to the information to be pushed Hobby value；If the hobby value calculated exceedes the hobby threshold value pre-set, treat that push is believed to described in target push Breath, information push quality can be lifted, to solve in the method for existing pushed information, be learnt what is obtained according to disassembler The problem of hobby difference in different time cycle is larger, and information push is of low quality.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing There is the required accompanying drawing used in technology description to be briefly described, it should be apparent that, drawings in the following description are only this Some embodiments of invention, for those of ordinary skill in the art, on the premise of not paying creative work, can be with Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is the method flow schematic diagram of the pushed information of embodiments of the invention one；
Fig. 2 is the method flow schematic diagram that embodiments of the invention one calculate characteristic statisticses vector；
Fig. 3 is the schematic diagram of the rotation renewal of embodiments of the invention one first memory block to the 3rd memory block；
Fig. 4 is the apparatus structure schematic diagram of the pushed information of embodiments of the invention two；
Fig. 5 is the structural representation of electronic equipment one embodiment of the present invention.
Embodiment
The embodiment of the present invention is described in detail below in conjunction with the accompanying drawings.
It will be appreciated that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.Base Embodiment in the present invention, those of ordinary skill in the art obtained under the premise of creative work is not made it is all its Its embodiment, belongs to the scope of protection of the invention.
Fig. 1 is the method flow schematic diagram of the pushed information of embodiments of the invention one, as shown in figure 1, the side of the present embodiment Method can include：
Step 101, learnt using disassembler, the learning characteristic vector in the multiple time cycles pre-set, often One learning characteristic vector answers the feature in information to be pushed；
In the present embodiment, the learning characteristic on obtaining information to be pushed using learning machine is known technology, is omitted here It is described in detail.
In the present embodiment, because disassembler (Factorization Machines) is a kind of implicit method, after study The every one-dimensional characteristic (characteristic vector) arrived all enters without definite implication, the data twice of different time sections (time cycle) input Go after learning, obtained characteristic vector non-equivalence.In the present embodiment, user characteristics is being regularly updated with the period (time cycle) To ensure outside to the seizure of user preferences, by considering multiple periods, in favor of excavating the more constant hobby of user, example Such as, it is based on study daily for prior art and obtains learning characteristic vector, is entered by accumulating statistic of the user characteristics in 2 days Row characteristic vector learns, and can be beneficial to the more constant hobby for excavating user.
Step 102, obtain the information to be pushed in the multiple time cycle browses statistical vector；
In the present embodiment, as an alternative embodiment, it can be that user browsing behavior is all in multiple times to browse statistic Real-time progressive in phase.For example, within two time cycles of accumulation, each user is counted respectively to being included in information to be pushed Browse (view), click (click), comment etc. per one-dimensional characteristic browse statistic.
Step 103, according to the learning characteristic it is vectorial and it is described browse statistical vector, calculate in the information to be pushed Comprising every one-dimensional characteristic characteristic statisticses vector；
In the present embodiment, as an alternative embodiment, characteristic statisticses vector is calculated using following formula：
In formula,
StatiFor the characteristic statisticses vector of the i-th dimension feature included in the information to be pushed；
frejWithin the multiple time cycle user of the information to be pushed is browsed for jth position user statistics to Amount；
featurejiFor jth position user within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed to Amount；
N is the number of users of statistics.
In the present embodiment, it is navigation patterns of the user to information to be pushed to browse statistical vector, for example, to information to be pushed Number of clicks, number of visits within the multiple time cycle, click on frequency, browse frequency etc..
Can also be that the learning characteristic of different dimensional is vectorial for same user as an alternative embodiment in the present embodiment Different characteristic vector coefficients is set.Therefore, it is possible to calculate characteristic statisticses vector using following formula：
In formula,
ξjiFor i-th dimension characteristic vector coefficient corresponding to the user of jth position.
Can also be that different user sets different characteristic statisticses vectors as another alternative embodiment in the present embodiment Coefficient.Therefore, it is possible to calculate characteristic statisticses vector using following formula：
In formula,
ζjFor the characteristic statisticses vector coefficient of jth position user.
Step 104, according to characteristic statisticses vector and the learning characteristic vector of target, the target is calculated to described The hobby value of information to be pushed；
In the present embodiment, as an alternative embodiment, happiness of the target to the information to be pushed is calculated using following formula Good value：
In formula,
ψ is hobby value of the target to information to be pushed；
featureuiFor target within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed vector；
K is the intrinsic dimensionality included in information to be pushed.
In the present embodiment, obtain new user using Factorization Machines and browse statistical vector and to information Characteristic statisticses vector, statistical vector can be browsed with real-time update user, can make up characteristic statisticses vector itself due to calculate provide Source limit and the defects of be unable to real-time update so that learnt by the data twice in the time cycle to accumulation, can be with Effectively reduce hobby deviation of the user based on disassembler technology to information.
Step 105, if the hobby value calculated exceedes the hobby threshold value pre-set, to described in target push Information to be pushed.
In the present embodiment, the hobby threshold value pre-set can be configured according to being actually needed.If the happiness calculated Good value exceedes the hobby threshold value pre-set, shows that the hobby of targeted customer more matches with information to be pushed, if push should Information, the probability that targeted customer browses the pushed information is higher, effectively improves the information quality of push, so as on the one hand can be with The applied business experience of user is lifted, on the other hand can also extend the value-added service application of related operators.
The method of the pushed information of embodiments of the invention one, learns by using disassembler, and what is pre-set is multiple Learning characteristic vector in time cycle, each learning characteristic vector answer the feature in information to be pushed；Obtain the multiple The information to be pushed in time cycle browses statistical vector；It is vectorial and described browse statistics according to the learning characteristic Vector, calculate the characteristic statisticses vector of the every one-dimensional characteristic included in the information to be pushed；According to characteristic statisticses vector And the learning characteristic vector of target, calculate hobby value of the target to the information to be pushed；If the happiness calculated Good value exceedes the hobby threshold value pre-set, and the information to be pushed is pushed to the target.So, during multiple based on accumulation Between computation of Period learning characteristic vector and browse statistical vector, the hobby value that can effectively reduce the information being calculated is inclined Difference, it can effectively lift the information quality of push.
Below so that the method for rotation renewal calculates characteristic statisticses vector as an example, the present embodiment is made further specifically It is bright.
Fig. 2 is the method flow schematic diagram that embodiments of the invention one calculate characteristic statisticses vector.As shown in Fig. 2 with pre- The multiple time cycles first set are 3 time cycles, are that the method for the present embodiment can wrap exemplified by one day per a period of time Include：
Step 201, user is obtained in today to the characteristic statisticses of information to be pushed vector and characteristic information and with first Memory block stores；
In the present embodiment, set the first memory block (DB1), storage is calculated today with Factorization Machines User concealed characteristic information and the characteristic statisticses vector arrived.
Step 202, at second day, the characteristic statisticses vector and characteristic information that were obtained to yesterday are updated and use second Memory block stores, and, user is obtained in today to the characteristic statisticses of information to be pushed vector and characteristic information and with first Memory block stores；
In the present embodiment, the first memory block (DB1) is renamed as the second memory block (DB2), to the yesterday after being updated The characteristic statisticses vector and characteristic information of acquisition are stored, i.e. counted user concealed feature yesterday of DB2 storage renewals Information and characteristic statisticses vector.
After being renamed to the first memory block (DB1), set the first memory block (DB1), today is used Factorization in storage User concealed characteristic information and the characteristic statisticses vector that Machines is calculated.
Step 203, at the 3rd day, the characteristic statisticses vector and characteristic information that were obtained to the day before yesterday are updated and use the 3rd Memory block stores, and the characteristic statisticses vector and characteristic information obtained to yesterday is updated and uses the second memory block to store, with And acquisition user is stored to the characteristic statisticses vector and characteristic information of information to be pushed and with the first memory block in today；
In the present embodiment, the second memory block (DB2) is renamed as the 3rd memory block (DB3), to the day before yesterday after being updated The characteristic statisticses vector and characteristic information of acquisition are stored, i.e. the day before yesterday counted user concealed feature of DB3 storage renewals Information and characteristic statisticses vector.
After being renamed to the second memory block (DB2), by the first memory block (DB1), the second memory block (DB2) is renamed as, it is right The characteristic statisticses vector and characteristic information obtained the yesterday after being updated is stored, i.e. the yesterday of DB2 storage renewals calculates User concealed characteristic information and the characteristic statisticses vector obtained.
After being renamed to the first memory block (DB1), set the first memory block (DB1), today is used Factorization in storage User concealed characteristic information and the characteristic statisticses vector that Machines is calculated.
In the present embodiment, for each time viewing of the user in today, using characteristic statisticses vector calculation formula, exist respectively The characteristic statisticses of every one-dimensional characteristic of the real-time update user to being included in the information to be pushed are vectorial in DB1, DB2, DB3, this Sample is accumulated simultaneously equivalent in 3 different characteristic statisticses in, can solve implication change before and after implicit features update and bring Cumulant Problem of Failure.
Step 204, characteristic statisticses vector renewal obtained to the day before yesterday is as the user being calculated to described The characteristic statisticses vector of the every one-dimensional characteristic included in information to be pushed；
In the present embodiment, because DB3 has already been through the accumulation of the two days day before yesterday/yesterdays, meet that the excavation that is beneficial to of two days is used The time requirement of family hobby, thus, calculate hobby value and use DB3.
Step 205, at the 4th day, empty the 3rd memory block, perform the characteristic statisticses vector obtained to the day before yesterday and The step of characteristic information is updated and uses the 3rd memory block to store.
Fig. 3 is the schematic diagram of the rotation renewal of embodiments of the invention one first memory block to the 3rd memory block.Such as Fig. 3 institutes Show, the characteristic statisticses vector and characteristic information stored in memory block is regularly updated using the method for rotation renewal.
In the present embodiment, at the end of today, DB1 is renamed as DB2, and DB2 is renamed as DB3, and DB3 is emptied and is renamed as DB1, Prepare the latest features statistical vector and characteristic information of accumulation tomorrow, change so as to solve implication before and after implicit features renewal The cumulant Problem of Failure brought so that implicit features can be integrated into content-based recommendation system.
Fig. 4 is the apparatus structure schematic diagram of the pushed information of embodiments of the invention two, as shown in figure 4, the dress of the present embodiment Putting to include：Learning characteristic vector acquisition module 41, browse statistical vector acquisition module 42, characteristic statisticses vector calculation module 43rd, hobby value computing module 44 and pushing module 45, wherein,
Learning characteristic vector acquisition module 41, for being learnt using disassembler, the multiple time cycles pre-set Interior learning characteristic vector, each learning characteristic vector answer the feature in information to be pushed；
Statistical vector acquisition module 42 is browsed, for obtaining the clear of the information to be pushed in the multiple time cycle Look at statistical vector；
In the present embodiment, as an alternative embodiment, it can be that user browsing behavior is all in multiple times to browse statistic Real-time progressive in phase.For example, within two time cycles of accumulation, each user is counted respectively to being included in information to be pushed Browsing, clicking on, commenting on etc. per one-dimensional characteristic browses statistic.
Characteristic statisticses vector calculation module 43, for according to the learning characteristic it is vectorial and it is described browse statistical vector, Calculate the characteristic statisticses vector of the every one-dimensional characteristic included in the information to be pushed；
In the present embodiment, as an alternative embodiment, the characteristic statisticses vector is calculated using following formula：
In formula,
StatiFor the characteristic statisticses vector of the i-th dimension feature included in the information to be pushed；
frejWithin the multiple time cycle user of the information to be pushed is browsed for jth position user statistics to Amount；
featurejiFor jth position user within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed to Amount；
N is the number of users of statistics.
As another alternative embodiment, following formula can also be utilized to calculate the characteristic statisticses vector：
In formula,
ζjiFor i-th dimension characteristic vector coefficient corresponding to the user of jth position；
ζjFor the characteristic statisticses vector coefficient of jth position user.
As yet another alternative embodiment, following formula can also be utilized to calculate the characteristic statisticses vector：
In the present embodiment, as yet another alternative embodiment, characteristic statisticses vector calculation module 43 includes：The cycle very first time Processing unit, the second time cycle processing unit, the 3rd time cycle processing unit, characteristic statisticses vector acquiring unit and wheel Turn unit (not shown), wherein,
Very first time period treatment unit, for obtaining the characteristic statisticses vector of the information to be pushed in the cycle very first time And characteristic information and with the first memory block store；
Second time cycle processing unit, within the second time cycle, to the feature obtained in the cycle very first time Statistical vector and characteristic information are updated and use the second memory block to store, and, obtain and wait to push away in the second time cycle Deliver letters breath characteristic statisticses vector and characteristic information and with the first memory block storage；
3rd time cycle processing unit, within the 3rd time cycle, to the feature obtained in the cycle very first time Statistical vector and characteristic information are updated and use the 3rd memory block to store, to the characteristic statisticses obtained in the second time cycle Vector and characteristic information are updated and use the second memory block to store, and, obtain in the 3rd time cycle and treat push letter The characteristic statisticses vector and characteristic information of breath are simultaneously stored with the first memory block；
Characteristic statisticses vector acquiring unit, for the week very first time that will be updated in the 3rd time cycle processing unit The characteristic statisticses vector obtained in phase is united as the feature of the every one-dimensional characteristic included in the information to be pushed being calculated Meter vector；
Unit is rotated, within the 4th time cycle, emptying the 3rd memory block, notifies the 3rd time cycle to handle Unit performs the characteristic statisticses vector and characteristic information to being obtained in the cycle very first time and is updated and uses the 3rd to deposit The step of storage area stores.
Hobby value computing module 44, for the learning characteristic vector according to characteristic statisticses vector and target, calculate Hobby value of the target to the information to be pushed；
In the present embodiment, as an alternative embodiment, happiness of the target to the information to be pushed is calculated using following formula Good value：
In formula,
ψ is hobby value of the target to information to be pushed；
featureuiFor target within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed vector；
K is the intrinsic dimensionality included in information to be pushed.
Pushing module 45, if the hobby value calculated exceedes the hobby threshold value pre-set, pushed to the target The information to be pushed.
The device of the present embodiment, it can be used for the technical scheme for performing embodiment of the method shown in Fig. 1 to Fig. 3, it realizes former Reason is similar with technique effect, and here is omitted.
It should be noted that herein, such as first and second or the like relational terms are used merely to a reality Body or operation make a distinction with another entity or operation, and not necessarily require or imply and deposited between these entities or operation In any this actual relation or order.Moreover, term " comprising ", "comprising" or its any other variant are intended to Nonexcludability includes, so that process, method, article or equipment including a series of elements not only will including those Element, but also the other element including being not expressly set out, or it is this process, method, article or equipment also to include Intrinsic key element.In the absence of more restrictions, the key element limited by sentence "including a ...", it is not excluded that Other identical element also be present in process, method, article or equipment including the key element.
Each embodiment in this specification is described by the way of related, identical similar portion between each embodiment Divide mutually referring to what each embodiment stressed is the difference with other embodiment.
For device embodiment, because it is substantially similar to embodiment of the method, so the comparison of description is simple Single, the relevent part can refer to the partial explaination of embodiments of method.
Expression or logic and/or step described otherwise above herein in flow charts, for example, being considered use In the order list for the executable instruction for realizing logic function, may be embodied in any computer-readable medium, for Instruction execution system, device or equipment (such as computer based system including the system of processor or other can be held from instruction The system of row system, device or equipment instruction fetch and execute instruction) use, or combine these instruction execution systems, device or set It is standby and use.For the purpose of this specification, " computer-readable medium " can any can be included, store, communicate, propagate or pass Defeated program is for instruction execution system, device or equipment or the dress used with reference to these instruction execution systems, device or equipment Put.The more specifically example (non-exhaustive list) of computer-readable medium includes following：Electricity with one or more wiring Connecting portion (electronic installation), portable computer diskette box (magnetic device), random access memory (RAM), read-only storage (ROM), erasable edit read-only storage (EPROM or flash memory), fiber device, and portable optic disk is read-only deposits Reservoir (CDROM).In addition, computer-readable medium, which can even is that, to print the paper of described program thereon or other are suitable Medium, because can then enter edlin, interpretation or if necessary with it for example by carrying out optical scanner to paper or other media His suitable method is handled electronically to obtain described program, is then stored in computer storage.
It should be appreciated that each several part of the present invention can be realized with hardware, software, firmware or combinations thereof.
In the above-described embodiment, multiple steps or method can use storage to be performed in memory and by suitable instruction The software or firmware that system performs are realized.If for example, being realized with hardware, with another embodiment, can use Any one of following technology well known in the art or their combination are realized：With for realizing logic work(to data-signal The discrete logic of the logic gates of energy, there is the application specific integrated circuit of suitable combinational logic gate circuit, programmable gate Array (PGA), field programmable gate array (FPGA) etc..
The embodiment of the present invention also provides a kind of electronic equipment, and the electronic equipment includes the dress described in foregoing any embodiment Put.
Fig. 5 is the structural representation of electronic equipment one embodiment of the present invention, it is possible to achieve is implemented shown in Fig. 1-4 of the present invention The flow of example, as shown in figure 5, above-mentioned electronic equipment can include：Housing 51, processor 52, memory 53, circuit board 54 and electricity Source circuit 55, wherein, circuit board 54 is placed in the interior volume that housing 51 surrounds, and processor 52 and memory 53 are arranged on circuit On plate 54；Power circuit 55, for each circuit or the device power supply for above-mentioned electronic equipment；Memory 53 is used to store and can hold Line program code；Processor 52 is run and executable program generation by reading the executable program code stored in memory 53 Program corresponding to code, the method for performing the pushed information described in foregoing any embodiment.
Processor 52 to the specific implementation procedures of above-mentioned steps and processor 52 by run executable program code come The step of further performing, the description of Fig. 1-4 illustrated embodiments of the present invention is may refer to, will not be repeated here.
The electronic equipment exists in a variety of forms, includes but is not limited to：
(1) mobile communication equipment：The characteristics of this kind equipment is that possess mobile communication function, and to provide speech, data Communicate as main target.This Terminal Type includes：Smart mobile phone (such as iPhone), multimedia handset, feature mobile phone, and it is low Hold mobile phone etc..
(2) super mobile personal computer equipment：This kind equipment belongs to the category of personal computer, there is calculating and processing work( Can, typically also possess mobile Internet access characteristic.This Terminal Type includes：PDA, MID and UMPC equipment etc., such as iPad.
(3) portable entertainment device：This kind equipment can show and play content of multimedia.The kind equipment includes：Audio, Video player (such as iPod), handheld device, e-book, and intelligent toy and portable car-mounted navigation equipment.
(4) server：The equipment for providing the service of calculating, the composition of server are total including processor, hard disk, internal memory, system Line etc., server is similar with general computer architecture, but due to needing to provide highly reliable service, therefore in processing energy Power, stability, reliability, security, scalability, manageability etc. require higher.
(5) other electronic equipments with data interaction function.
Those skilled in the art are appreciated that to realize all or part of step that above-described embodiment method carries Suddenly it is that by program the hardware of correlation can be instructed to complete, described program can be stored in a kind of computer-readable storage medium In matter, the program upon execution, including one or a combination set of the step of embodiment of the method.
For convenience of description, it is to be divided into various units/modules with function to describe respectively to describe apparatus above.Certainly, exist The function of each unit/module can be realized in same or multiple softwares and/or hardware when implementing of the invention.
As seen through the above description of the embodiments, those skilled in the art can be understood that this
Invention can add the mode of required general hardware platform to realize by software.Based on such understanding, the present invention The part that is substantially contributed in other words to prior art of technical scheme can be embodied in the form of software product, should Computer software product can be stored in storage medium, such as ROM/RAM, magnetic disc, CD, including some instructions are causing One computer equipment (can be personal computer, server, or network equipment etc.) perform each embodiment of the present invention or Method described in some parts of person's embodiment.
The foregoing is only a specific embodiment of the invention, but protection scope of the present invention is not limited thereto, any Those familiar with the art the invention discloses technical scope in, the change or replacement that can readily occur in, all should It is included within the scope of the present invention.Therefore, protection scope of the present invention should be defined by scope of the claims.

Claims (10)

  1. A kind of 1. method of pushed information, it is characterised in that including：
    Learnt using disassembler, learning characteristic in the multiple time cycles pre-set vector, each learning characteristic to Amount answers the feature in information to be pushed；
    Obtain the information to be pushed in the multiple time cycle browses statistical vector；
    According to the learning characteristic it is vectorial and it is described browse statistical vector, calculate included in the information to be pushed it is every one-dimensional The characteristic statisticses vector of feature；
    According to characteristic statisticses vector and the learning characteristic vector of target, the target is calculated to the information to be pushed Hobby value；
    If the hobby value calculated exceedes the hobby threshold value pre-set, the information to be pushed is pushed to the target.
  2. 2. the method for pushed information according to claim 1, it is characterised in that using following formula calculate the characteristic statisticses to Amount：
    <mrow> <msub> <mi>Stat</mi> <mi>i</mi> </msub> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <msub> <mi>fre</mi> <mi>j</mi> </msub> <msub> <mi>xfeature</mi> <mrow> <mi>j</mi> <mi>i</mi> </mrow> </msub> </mrow>
    In formula,
    StatiFor the characteristic statisticses vector of the i-th dimension feature included in the information to be pushed；
    frejStatistical vector is browsed to the information to be pushed within the multiple time cycle for jth position user；
    featurejiFor jth position user within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed vector；
    N is the number of users of statistics.
  3. 3. the method for pushed information according to claim 1, it is characterised in that using following formula calculate the characteristic statisticses to Amount：
    <mrow> <msub> <mi>Stat</mi> <mi>i</mi> </msub> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <msub> <mi>&amp;zeta;</mi> <mi>j</mi> </msub> <msub> <mi>xfre</mi> <mi>j</mi> </msub> <msub> <mi>x&amp;xi;</mi> <mrow> <mi>j</mi> <mi>i</mi> </mrow> </msub> <msub> <mi>xfeature</mi> <mrow> <mi>j</mi> <mi>i</mi> </mrow> </msub> </mrow>
    In formula,
    ξjiFor i-th dimension characteristic vector coefficient corresponding to the user of jth position；
    ζjFor the characteristic statisticses vector coefficient of jth position user.
  4. 4. the method for the pushed information according to Claims 2 or 3, it is characterised in that calculate the target pair using following formula The hobby value of the information to be pushed：
    <mrow> <mi>&amp;psi;</mi> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>k</mi> </munderover> <msub> <mi>Stat</mi> <mi>i</mi> </msub> <msub> <mi>xfeature</mi> <mrow> <mi>u</mi> <mi>i</mi> </mrow> </msub> </mrow>
    In formula,
    ψ is hobby value of the target to information to be pushed；
    featureuiFor target within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed vector；
    K is the intrinsic dimensionality included in information to be pushed.
  5. 5. the method for pushed information according to claim 1, it is characterised in that described calculate is wrapped in the information to be pushed The characteristic statisticses vector of the every one-dimensional characteristic contained includes：
    The characteristic statisticses vector and characteristic information of information to be pushed in the acquisition cycle very first time are simultaneously deposited with the first memory block Storage；
    Within the second time cycle, the characteristic statisticses vector and characteristic information obtained in the cycle very first time is updated simultaneously Stored with the second memory block, and, the characteristic statisticses vector and feature for obtaining the information to be pushed in the second time cycle are believed Cease and use the first memory block to store；
    Within the 3rd time cycle, the characteristic statisticses vector and characteristic information obtained in the cycle very first time is updated simultaneously Stored with the 3rd memory block, the characteristic statisticses vector and characteristic information obtained in the second time cycle is updated and uses the Two memory blocks store, and, the characteristic statisticses vector and characteristic information of the information to be pushed in the 3rd time cycle of acquisition are simultaneously Stored with the first memory block；
    Using the characteristic statisticses obtained in the cycle very first time of renewal vector as being wrapped in the information to be pushed being calculated The characteristic statisticses vector of the every one-dimensional characteristic contained；
    Within the 4th time cycle, empty the 3rd memory block, perform it is described to the characteristic statisticses that are obtained in the cycle very first time to The step of amount and characteristic information are updated and use the 3rd memory block to store.
  6. A kind of 6. device of pushed information, it is characterised in that including：Learning characteristic vector acquisition module, browse statistical vector and obtain Modulus block, characteristic statisticses vector calculation module, hobby value computing module and pushing module, wherein,
    Learning characteristic vector acquisition module, for being learnt using disassembler, in multiple time cycles pre-set Characteristic vector is practised, each learning characteristic vector answers the feature in information to be pushed；
    Statistical vector acquisition module is browsed, statistics is browsed for obtain the information to be pushed in the multiple time cycle Vector；
    Characteristic statisticses vector calculation module, for according to the learning characteristic it is vectorial and it is described browse statistical vector, calculate institute State the characteristic statisticses vector of the every one-dimensional characteristic included in information to be pushed；
    Hobby value computing module, for the learning characteristic vector according to characteristic statisticses vector and target, calculate the mesh Mark the hobby value to the information to be pushed；
    Pushing module, if the hobby value calculated exceedes the hobby threshold value pre-set, treated to described in target push Pushed information.
  7. 7. the device of pushed information according to claim 6, it is characterised in that using following formula calculate the characteristic statisticses to Amount：
    <mrow> <msub> <mi>Stat</mi> <mi>i</mi> </msub> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <msub> <mi>fre</mi> <mi>j</mi> </msub> <msub> <mi>xfeature</mi> <mrow> <mi>j</mi> <mi>i</mi> </mrow> </msub> </mrow>
    In formula,
    StatiFor the characteristic statisticses vector of the i-th dimension feature included in the information to be pushed；
    frejStatistical vector is browsed to the user of the information to be pushed within the multiple time cycle for jth position user；
    featurejiFor jth position user within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed vector；
    N is the number of users of statistics.
  8. 8. the device of pushed information according to claim 6, it is characterised in that using following formula calculate the characteristic statisticses to Amount：
    <mrow> <msub> <mi>Stat</mi> <mi>i</mi> </msub> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>j</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>N</mi> </munderover> <msub> <mi>&amp;zeta;</mi> <mi>j</mi> </msub> <msub> <mi>xfre</mi> <mi>j</mi> </msub> <msub> <mi>x&amp;xi;</mi> <mrow> <mi>j</mi> <mi>i</mi> </mrow> </msub> <msub> <mi>xfeature</mi> <mrow> <mi>j</mi> <mi>i</mi> </mrow> </msub> </mrow>
    In formula,
    ξjiFor i-th dimension characteristic vector coefficient corresponding to the user of jth position；
    ζjFor the characteristic statisticses vector coefficient of jth position user.
  9. 9. the device of the pushed information according to claim 7 or 8, it is characterised in that calculate the target pair using following formula The hobby value of the information to be pushed：
    <mrow> <mi>&amp;psi;</mi> <mo>=</mo> <munderover> <mo>&amp;Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>k</mi> </munderover> <msub> <mi>Stat</mi> <mi>i</mi> </msub> <msub> <mi>xfeature</mi> <mrow> <mi>u</mi> <mi>i</mi> </mrow> </msub> </mrow>
    In formula,
    ψ is hobby value of the target to information to be pushed；
    featureuiFor target within the multiple time cycle to the i-th dimension learning characteristic of information to be pushed vector；
    K is the intrinsic dimensionality included in information to be pushed.
  10. 10. the device of pushed information according to claim 6, it is characterised in that the characteristic statisticses vector calculation module Including：Very first time period treatment unit, the second time cycle processing unit, the 3rd time cycle processing unit, characteristic statisticses Vectorial acquiring unit and rotation unit, wherein,
    Very first time period treatment unit, for obtain the information to be pushed in the cycle very first time characteristic statisticses vector and Characteristic information is simultaneously stored with the first memory block；
    Second time cycle processing unit, within the second time cycle, to the characteristic statisticses obtained in the cycle very first time Vector and characteristic information are updated and use the second memory block to store, and, obtain in the second time cycle and treat push letter The characteristic statisticses vector and characteristic information of breath are simultaneously stored with the first memory block；
    3rd time cycle processing unit, within the 3rd time cycle, to the characteristic statisticses obtained in the cycle very first time Vector and characteristic information are updated and use the 3rd memory block to store, to the characteristic statisticses vector obtained in the second time cycle And characteristic information is updated and uses the second memory block to store, and, obtain the information to be pushed in the 3rd time cycle Characteristic statisticses vector and characteristic information are simultaneously stored with the first memory block；
    Characteristic statisticses vector acquiring unit, in cycle very first time for will being updated in the 3rd time cycle processing unit Acquisition characteristic statisticses vector as the every one-dimensional characteristic included in the information to be pushed being calculated characteristic statisticses to Amount；
    Unit is rotated, within the 4th time cycle, emptying the 3rd memory block, notifies the 3rd time cycle processing unit The characteristic statisticses vector and characteristic information to obtaining in the cycle very first time is performed to be updated and use the 3rd memory block The step of storage.
CN201610513018.0A 2016-06-30 2016-06-30 Information pushing method and device and electronic equipment Active CN107562758B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610513018.0A CN107562758B (en) 2016-06-30 2016-06-30 Information pushing method and device and electronic equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610513018.0A CN107562758B (en) 2016-06-30 2016-06-30 Information pushing method and device and electronic equipment

Publications (2)

Publication Number Publication Date
CN107562758A true CN107562758A (en) 2018-01-09
CN107562758B CN107562758B (en) 2020-12-01

Family

ID=60969169

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610513018.0A Active CN107562758B (en) 2016-06-30 2016-06-30 Information pushing method and device and electronic equipment

Country Status (1)

Country Link
CN (1) CN107562758B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109033460A (en) * 2018-08-30 2018-12-18 优视科技新加坡有限公司 Sort method, device and equipment/terminal/server in a kind of information flow
CN109617999A (en) * 2019-01-08 2019-04-12 平安科技（深圳）有限公司 Content delivery method, device, computer equipment and storage medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102982107A (en) * 2012-11-08 2013-03-20 北京航空航天大学 Recommendation system optimization method with information of user and item and context attribute integrated
CN104615790A (en) * 2015-03-09 2015-05-13 百度在线网络技术（北京）有限公司 Characteristic recommendation method and device
CN105117418A (en) * 2015-07-30 2015-12-02 百度在线网络技术（北京）有限公司 Search based service information management system and method
US20160132601A1 (en) * 2014-11-12 2016-05-12 Microsoft Technology Licensing Hybrid Explanations In Collaborative Filter Based Recommendation System

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102982107A (en) * 2012-11-08 2013-03-20 北京航空航天大学 Recommendation system optimization method with information of user and item and context attribute integrated
US20160132601A1 (en) * 2014-11-12 2016-05-12 Microsoft Technology Licensing Hybrid Explanations In Collaborative Filter Based Recommendation System
CN104615790A (en) * 2015-03-09 2015-05-13 百度在线网络技术（北京）有限公司 Characteristic recommendation method and device
CN105117418A (en) * 2015-07-30 2015-12-02 百度在线网络技术（北京）有限公司 Search based service information management system and method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
雪伦..: "因子分解机FM原理及SGD训练", 《网络公开 HTTPS://BLOG.CSDN.NET/A819825294/ARTICLE/DETAILS/51218296?LOCATIONNUM=13》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109033460A (en) * 2018-08-30 2018-12-18 优视科技新加坡有限公司 Sort method, device and equipment/terminal/server in a kind of information flow
CN109617999A (en) * 2019-01-08 2019-04-12 平安科技（深圳）有限公司 Content delivery method, device, computer equipment and storage medium

Also Published As

Publication number Publication date
CN107562758B (en) 2020-12-01

Similar Documents

Publication Publication Date Title
Hidano et al. Model inversion attacks for prediction systems: Without knowledge of non-sensitive attributes
Sala et al. Sharing graphs using differentially private graph models
CN102890689B (en) The method for building up and system of a kind of user interest model
CN104765751B (en) Using recommended method and device
CN110503206A (en) A kind of prediction model update method, device, equipment and readable medium
CN107169010A (en) A kind of determination method and device of recommendation search keyword
CN105426365B (en) Distinguish the method and device of interbehavior
CN106649236B (en) Modify the method and device of prompt
CN104246751A (en) Recommendation device, recommendation system, recommendation method and program
CN109388674A (en) Data processing method, device, equipment and readable storage medium storing program for executing
CN108038229A (en) Government affairs information search method, system and terminal device
CN110110226A (en) A kind of proposed algorithm, recommender system and terminal device
CN109977316A (en) A kind of parallel type article recommended method, device, equipment and storage medium
CN106803092B (en) Method and device for determining standard problem data
CN107562758A (en) Information pushing method and device and electronic equipment
Razghandi et al. A context-aware and user behavior-based recommender system with regarding social network analysis
Nikolaou et al. The bag semantics of ontology-based data access
CN112116426A (en) Method and device for pushing article information
Klimova et al. Iterative scaling in curved exponential families
CN108763260A (en) A kind of examination question searching method, system and terminal device
CN110516033A (en) A kind of method and apparatus calculating user preference
Sharma et al. Recommender System: A bibliometric analysis
CN109450963A (en) Information push method and terminal device
CN108319721A (en) Application program recommends method, apparatus, user terminal and server
Hazarika Pointwise ideal convergence and uniformly ideal convergence of sequences of fuzzy valued functions

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant